sEVs were isolated from equal number of cells in each of the different muscle cell states after medium was conditioned for an equal time interval (6 h, Figure 1A). Given that the different cell states require distinct culture conditions, appropriate exosome production medium (EPM) depleted of serum exosomes was generated for each culture state (see Methods) and tested to ensure equivalent growth promoting activity as control media (Supplementary Figure S2). Thus, our protocol ensures collection of sEVs secreted in a defined time by a defined number of cells in distinct physiologically relevant stable states.

Unexpectedly, G₀ cells secreted substantially more sEVs than an equivalent number of proliferating MB, as evidenced by the yield of total protein in exosome fraction (Figure 1B) as well as by nanoparticle tracking analysis (NTA) (Figure 1C). This enhanced secretion was reversible: 24 h after G₀ cells were induced to re-enter the cell cycle (R24), sEV secretion returned to the levels seen in asynchronously proliferating MB. Secretion rate in terminally arrested MT was also similar to that in MB. Thus, despite a lower overall metabolic rate (Valcourt et al., 2012), G₀ cells unexpectedly showed enhanced sEV secretion compared to other cellular states. In all four states, particle size (mean diameter) estimated by NTA (140–180 nm, Figure 1D, Supplementary Figure S3A,B) was consistent with that reported for exosomes (Dehghani et al., 2020).

To confirm the composition of the isolated fraction, GW4869 (2.5 µM) an inhibitor of neutral sphingomyelinase was used to directly inhibit exosome biogenesis, leading to ∼50% reduction in sEV particles in CM (Figure 1E). TEM analysis confirmed that size of purified sEV from all states was in the 150 nm range (Figure 1F). Taken together, these observations suggest an altered regulation of MVB biogenesis and/or sEV release during quiescence, which was reversed as cells re-entered proliferation.

Given the significant increase in sEV output in G₀, we further characterized these vesicles. We confirmed that the fraction purified by ultracentrifugation was enriched for exosomes using known markers: ESCRT-dependent early endosome proteins Alix and TSG101, lipid-raft protein flotillin, and tetraspanin CD9. Induction of quiescence for 6, 12 and 24 h did not alter the levels of cell-associated Alix, TSG101 and flotillin significantly (Figure 2A cell). However, G₀ cells (at 24 h) showed significant enrichment in exosome markers compared to MB, consistent with increased vesicle secretion (Figure 2B exo). Cell-associated levels of these markers did not vary much with cell state transitions, while sEV fractions showed variations consistent with changes in overall secretion rate and relative enrichment (Figures 2C–E). Compared to the cell lysate (Figure 2D), expression of tetraspanin CD9 was significantly increased in secreted G₀ sEV (Figures 2E,F) The sEV fractions from all states were negative for ER marker calnexin, confirming the exosomal enrichment. Taken together, the results indicate that quiescent cells show increased secretion of sEVs and enhanced exosome marker expression.

To confirm our results, we used sucrose density gradient centrifugation to separate and enrich secreted vesicles (Colombo et al., 2014) (Supplementary Figure S3C). TEM of pooled sucrose gradient fractions enriched for the sEV markers confirmed the enrichment of vesicles in the size range (∼150 nm diameter) expected for exosomes (Supplementary Figure S3D). The distribution of markers Alix and TSG101 in different fractions obtained from G₀, MB and MT were analyzed (Supplementary Figure S3E). For G₀, exosome markers were enriched in density fractions between 1.12 and 1.14 g/mL, while for MB these markers were enriched between 1.16 and 1.19 g/mL and for MT between 1.10 and 1.16 g/mL. These observations are consistent with differences in exosome cargo density and composition from different cellular states. Absence of calnexin confirmed absence of ER-derived vesicles.

Taken together, we provide evidence that quiescent myoblasts in culture secrete sEVs with exosome properties in higher quantities than proliferating or differentiated cells, and that cell cycle reactivation reverses this increased secretion, suggesting that exosome biogenesis/secretion represents a regulatory node during reversible cell cycle exit.

To explore the possible mechanisms behind increased sEV secretion in G₀, we investigated lipid droplets (LD), which provide a high degree of metabolic flexibility in skeletal muscle (Tan et al., 2021). The status of LD/lipid intermediates in G₀ myoblasts has not been reported. LD detected by Oil Red O (ORO) (Supplementary Figure S4A) were measured by quantitative imaging (Supplementary Figure S4B). MT showed 2.3-fold higher accumulation of LD as reported (Tan et al., 2021). Notably, we found that G₀ cells showed ∼3-fold more LD than MB, consistent with altered lipid metabolism in quiescence. Further, the distribution of ceramide-rich membranes (includes MVBs, ER and Golgi) using BODIPY^TR ceramide staining (Carayon et al., 2011; Horbay et al., 2022), showed profound architectural changes in G₀ compared to MB (Supplementary Figure S4C): high-resolution Airy scanning of a single confocal slice revealed numerous vesicles distributed to the periphery in G₀, whereas scattered fibrillar staining dominated in MB. Conceivably, alterations in LD and ceramide-rich intracellular membranes in G₀ may be involved not only in fatty acid oxidation which sustains survival in the low metabolic state of quiescence, but potentially for increased sEV biogenesis.

The scaffold protein Kibra has been implicated in regulating MVB biogenesis in neuronal cells (Song et al., 2019; Han et al., 2022). To investigate whether Kibra participates in regulation of exosome secretion in skeletal muscle cells, we measured endogenous (cell-associated) Kibra protein. We found that G₀ cells expressed significantly more Kibra protein than MB (Figures 3A,B), which was reversed as cells re-entered the cell cycle during reactivation, correlating with exosome secretion.

In neurons, Kibra functions by interacting with Rab27a, a key regulator of exosome biogenesis during ILV formation and MVB docking to PM (Song et al., 2019). G₀ cells expressed significantly more Rab27a protein than MB (Figures 3A,B), which like Kibra, was reversed during reactivation, and correlated with alterations in exosome secretion. Notably, the early endosome marker Eea1 also showed enhanced expression in G₀ (Figures 3C,D), consistent with increased MVB-exosome biogenesis during quiescence, since MVBs are of endosomal origin.

To assess whether Kibra and Rab27a proteins interact in myogenic cells as in neurons, we over-expressed both GFP-Kibra and Flag-mCherry-Rab27a in MB. Kibra was pulled down using anti-GFP nanobodies and its presence in GFP-Kibra:Flag-mCherryRab27a protein complex (124 kD, top panel) confirmed by immuno-blotting with anti-GFP (Figure 3E). Further, probing with Flag antibodies confirmed the enrichment of Flag-Rab27 (54 kD, lower panel, Figure 3E) confirming interaction of Kibra-GFP and Rab27a-flag-mCherry. Immuno-localization showed that Rab27a-mCherry co-localized with Kibra-GFP (Pearson’s correlation coefficient 0.76) (Figure 3F). Further, bonafide exosome markers CD63-GFP, CD9-GFP and YFP-Endo (YFP-tagged RhoB GTPase that marks endosome/MVB compartment), all showed considerable colocalization with Kibra/Rab27a in myoblasts (Figure 3G). Taken together, the results indicate a generalized increase in MVB biogenesis in G₀.

To determine whether Kibra affects sEV secretion rate, we over-expressed GFP-tagged Kibra in MB (Figure 4A). Ectopic expression of GFP-Kibra significantly increased the amount of protein in the exosome fraction by 3-fold over GFP control (Figure 4B), and NTA confirmed 30% increase in exosome numbers in Kibra-overexpressing MB (Figure 4C; Supplementary Figure S5A), along with enhanced expression of exosome markers Alix, Tsg101 and Hsp70 (∼3, 2.5 and 2-fold respectively, Figures 4D,E) in the purified sEV fraction. Thus, Kibra appears to enhance sEV secretion in muscle cells.

To further confirm the involvement of Kibra in sEV release, we inhibited endogenous Kibra expression in MB, using two distinct shRNAs (Figures 4F,G). Suppression of Kibra expression reduced sEV secretion to ∼40% as analysed by NTA, compared to cells expressing a control non-targeting sequence (Figure 4H; Supplementary Figure S5B), and decreased enrichment of exosome markers in the sEV fraction (Figures 4I,J), consistent with a role for Kibra in exosome release. Consistent with the reported role of Kibra in neurons, our results indicate a causal link between Kibra and sEV secretion in skeletal muscle cells as well, where increased expression of Kibra in G₀ cells is likely to facilitate the associated increase in sEV secretion.

To evaluate whether sEVs released by different cell states showed altered protein profiles we used label-free quantification of LC-MS/MS analysis (Figure 5). A total of 1,289 proteins were reproducibly detected in purified sEVs, with 233 proteins common to three states (MB, G₀, MT) (Figure 5A). Comparison of the 1,289 collated muscle cell sEV proteins with Exocarta (Keerthikumar et al., 2016) and Vesiclepedia (Pathan et al., 2019) databases (Figure 5B) revealed that >40% of known exosomal proteins including canonical exosomal markers TSG101, CD63, CD9, and CD81 and Alix were enriched, consistent with reproducible isolation of the exosome subclass of sEVs. Notably, the Wnt pathway effector β-catenin (MacDonald et al., 2009) was also found in sEVs from all three states, but there were quantitative differences in β-cat abundance (Mt>G₀>Mb), and MT sEV also carried Wnt10b and Wnt co-receptors Lrp1 and 4.

A comparison with exosome proteins reported from proliferating and differentiated muscle cells (Forterre et al., 2014) revealed substantial overlap (including Haspa8, Ncam, CD97, Cdh13, Col6a1, Col6a2, Tollip, Lgsals3, and Lgals3bp). As our study also included quiescent myoblast-derived exosomes which have not been previously reported, we looked for proteins selectively enriched in G₀ compared to MT (Figure 5C) and in G₀ compared to MB (Figure 5D). The volcano plots and gene ontology analysis highlight the proteins and pathways enriched in these binary comparisons, which together revealed that a signature of glutathione redox-related proteins (GSTA1, GSTP1, Gpx1, and Mpo) was found to be enriched in G₀-derived exosomes.

To determine if quiescence per se is associated with a particular protein signature, we compared proteomic data from quiescent neural stem cell (qNSC)-derived exosomes (Zhang et al., 2021): despite differences due to cell type, 202 proteins were commonly enriched between qNSC and quiescent (G₀) myoblast-derived exosome profiles (Figure 5E). String network analysis (https://string-db.org, version 12.0) revealed that 198/202 of these quiescence-enriched exosomal proteins showed direct protein-protein interactions (Figure 5F). In addition to the enrichment of ribosomal and proteosomal subunits and the chaperonin T-complex, of particular interest are transmembrane signaling proteins (integrins, CD44-Hyalunoran receptor, Mannose phosphate receptor), ECM components (Laminin, Fibronectin, HSPGs), intracellular signaling components (Cdc42, Rhog, and Rabs), and signaling enzymes (quiescin Qsox1 sulfhydryl transferase, PP2A, MAPK1). Immunoblotting of Qsox1 validated the differential enrichment in G₀ revealed by the proteomic analysis (Figure 5G). Thus, the quiescent state is associated with enrichment of proteins consistent with altered cargo loaded, suggesting possible differences in uptake and/or signaling capability of exosomes derived from different cell states. Overall, the proteomic analysis confirms that muscle cell derived sEV profiles substantially overlap with those of bona fide exosomes, that different muscle cell states are marked by both qualitative and quantitative changes in sEV profiles, and that sEVs from quiescent myoblasts share a common signature with quiescent NSCs.

Taken together, four features characterise the sEV fraction analysed thus far: (i) particle size and density (ii) exosome marker proteomic profile (iii) sensitivity to GW4869 a known inhibitor of exosome biogenesis (iv) sensitivity to perturbation of Kibra, a known regulator of exosome biogenesis. Therefore, we conclude that the sEV fraction under analysis is enriched for exosomes, and henceforth refer to these sEVs as exosomes.

Given the detection of Wnt pathway components in the proteomic analysis, we tested the functional capacity of muscle cell-derived exosomes to activate Wnt signaling in target cells. Exosomes purified from four different states were tested for their ability to activate an integrated TCF-Lef transcriptional reporter TOPflash in stable Wnt-responsive C2C12 cells (Venugopal et al., 2020). The small molecule Wnt activator, CHIR99201 served as a positive control and was used as an index of activation, inducing TOPflash activity ∼30-fold and ∼55-fold induction at 3 and 5 µM respectively (Figure 6A); the mutant FOPflash showed basal luciferase activity across samples. Equal amounts of exosomes (25 µg) derived from MB, G₀ and MT induced significant TOPflash activity in the same range as the chemical inducer [MB 20-fold, G₀ 27-fold, MT 42-fold], whereas R24-derived exosomes were less effective (∼14-fold activation) which can be represented as MT>G₀>MB > R24. Thus, exosomes derived from different cellular states elicited differential Wnt signaling response, consistent with the proteomics data showing that they differ in their signaling payload (specifically βcat and Wnts).

Exosome-mediated signaling may be affected by factors such as extent and mode of uptake by target cells. We measured the uptake of PKH26-labelled exosomes derived from different donor cell states (MB, G₀, R24 and MT) by target cells in a single state (proliferating MB). Donor exosomes (25 μg) were added at 37°C for 4 h followed by visualization and quantitatve fluorescence microscopy; cell boundaries and nuclei were visualized by staining with phalloidin and DAPI respectively (Figure 6B). Labelled exosomes from all states were actively taken up by MB (inhibited at 4°C, not shown), but to different extents. G₀-derived exosomes were least efficiently taken up by MB target cells (Figure 6C), MB and R24-derived exosomes to relatively equivalent levels, and MT-derived exosomes showed the strongest uptake by MB (summarized as MT>G₀>MB>R24). Notably, the extent of uptake did not correlate with the relative ability of exosomes from these states to activate Wnt signaling (MT>G₀>MB > R24), suggesting that other parameters such as the route or mechanism of uptake may regulate the final signaling outcome.

To determine whether exosomes derived from different donor cell states are internalized by different pathways, we used known inhibitors: heparin (competitive inhibitor of cell-surface HSPG-mediated uptake), wortmannin (PI3K inhibitor, disrupts macro-pinosome mediated uptake) and dynole (dynamin inhibitor, disrupts clathrin-mediated endocytic uptake). MB target cells were pre-incubated for 30 min with inhibitors [heparin (20 μg/mL), wortmannin (0.5 µM) or dynole (10 µM)], followed by co-incubation with PKH26-labelled exosomes (25 µg) as described. Consistent with reports that HSPGs participate in exosome uptake, heparin treatment of target cells reduced autocrine/self exosome internalization in all states (Supplementary Figure S6A–C). However, uptake of different donor state-derived exosomes by MB target cells showed differential sensitivity (Figure 6D). G₀-derived exosomes were internalized predominantly by HSPG-independent, clathrin-dependent pathways, MB and R24-derived donor exosomes were internalized via both HSPG-dependent and clathrin-dependent pathways, and MT-derived exosomes were taken up by HSPG-dependent, clathrin-dependent as well as wortmannin-sensitive pathways. The differential inhibitor sensitivity of uptake by target cells in a single state suggests distinct mechanisms for taking up different donor exosomes. Conceivably, the differential enrichment of integrins, ECM molecules, HSPGs and related proteins in different donor cell states (Figure 5) may play a role. Notably, uptake of G₀-derived exosomes was HSPG independent.

In skeletal muscle, differentiated myofibers far outnumber the G₀ MuSCs which represent only 1%–3% of nuclei in the tissue and reside in a myofiber niche. Following damage, to regenerate lost tissue, G₀ MuSCs are activated and give rise to large numbers of proliferating myoblasts. To model the possible communication between these components of muscle using our culture system, we next tested whether exosome uptake between specific donor-target pairs exhibit differential inhibitor sensitivity, using the assay described above.

MB and G ₀ donor-target pair not sensitive to HSPG inhibition: Whereas dynole significantly inhibited the uptake of G₀-derived exosomes by MB target cells, (Figure 6Dii, Figure 6E), heparin and wortmannin did not. Similarly, when MB-derived exosomes were added to G₀ target cells (Figures 6E,F), neither heparin nor wortmannin affected uptake significantly, but dynole did. Thus, exosome-dependent cross talk between the donor-target pair of G₀ and MB-cell states exhibited clathrin-dependent uptake, but is HSPG-independent.

G ₀ and MT donor-target pair sensitive to HSPG inhibition: We similarly assessed inhibitor sensitivity of exosome uptake between G₀ and MT donor-target pair (Figure 6G). MT-derived exosomes were internalized effectively by G₀ cells, but in addition to clathrin, uptake was heparin-sensitive. Similarly, internalization of G₀-derived exosomes by MT was sensitive to heparin. We conclude that internalization of exosomes in G₀ and MT employs both HSPG-mediated and clathrin-dependent endocytosis pathways, but as distinct from the MB-G₀ pair, uptake in between MT and G₀ was HSPG-dependent.

G ₀ exosomes are most efficiently internalized by MT: We assessed relative uptake of G₀ exosomes by MB vs. G₀ vs MT, considering uptake as PKH26 fluorescence intensity per unit area. G₀ exosomes were taken up ∼6-fold less efficiently by proliferating MB (Figure 6H, C.M.I, ∼1 × 10⁶) than differentiated MT (Figure 6I, C.M.I 6 × 10⁶), and at least 3-fold less efficiently than by G₀ target cells (Figure 6C, C.M.I. ∼0.7 × 10⁶). G₀ exosomes are enriched for HSPG2 expression (Figure 5), yet internalization of G₀ exosomes by MT but not by MB is heparin-sensitive (see above), implicating a target cell-dependent mechanism. Since HSPG abundance and glycation diversity are known to increase during muscle differentiation (Ghadiali et al., 2017; Mii and Takada, 2020), we infer that HSPGs on MT may be involved in the higher efficiency of uptake. Overall, the differential sensitivity and extent of uptake of exosomes suggest differences in exosome-mediated crosstalk between muscle cells in different states, with possible directionality of signaling, and differential utilization of HSPG pathways (Figure 6J).

Taken together, the experiments thus far provide evidence of state-dependent heterogeneity of exosomes in terms of quantitiative secretion, proteomic content, extent and pathways of uptake, and Wnt signaling function which is likely an integrated outcome of all of these aspects. In particular, G₀ exosomes are more efficiently taken up by MT than MB or G₀, via an HSPG-dependent pathway.

Given the heterogeneity in cargo and uptake of exosomes derived from different cell states, as well as signaling function with respect to a Wnt reporter, differences in target cell phenotypes may be expected. To determine whether donor exosomes derived from different cellular states elicit distinct phenotypic effects in a single target cell state, target MB were treated for 24 h with donor exosomes from MB, G₀, R24 or MT, and then assessed for proliferation and differentiation status. While treatment of proliferating MB with exosomes derived from MB, R24 or MT exosomes did not affect EdU incorporation significantly compared to untreated control, G₀ exosomes suppressed DNA synthesis (Figure 7A). Further, both G₀ and MT exosomes suppressed colony forming ability (Figure 7B), suggesting that they may carry inhibitors of proliferation and/or self-renewal.

Notably, treatment of proliferating MB with exosomes derived from G₀ cells induced differentiation, while neither MB nor MT-derived exosomes had this effect. G₀ exosomes uniquely induced Myogenin expression, as measured by increased frequency of Myogenin-expressing cells (Figures 7C,D), increased intensity of Myogenin expression per cell (Figure 7E), and enhanced Myogenin promoter-luciferase activity (Figure 7F). Further, exposure to G₀-exosomes led to multinucleated myotube formation with a 4-fold increase in fusion index and strong MyHC expression (Figures 7G,H), accompanied by 14-fold increase in muscle creatine kinase (MCK) promoter-luciferase activity (Figure 7I). Together, these results show that only G₀ exosomes are pro-myogenic, inducing the entire myogenic cascade, and overt phenotypic fusion into myotubes. Wnt signaling is known to be pro-myogenic (Tanaka et al., 2011; von Maltzahn et al., 2012; Subramaniam et al., 2014; Aloysius et al., 2018). Despite greater enrichment of Wnt signaling components including βcat, Wnt10, Lrp1/4 in MT-derived exosomes, only G₀-derived exosomes were able to collectively suppress proliferation, reduce self-renewal and induce differentiation in MB target cells. Based on the inhibitor sensitivity profile, G₀-derived exosomes may interact with target MB via an HSPG-independent mechanism to exert these phenotypic effects. Taken together, our data suggest that integrated signaling (including pathways other than Wnt) triggered by exosome binding/uptake may be responsible for the uniquely promyogenic functions of exosomes from G₀ cells.

Overall, this study reveals that proliferative and differentiation status of a single celltype influences secretory output as well as differential uptake of exosomes. This heterogeneity suggests the potential for altered signaling crosstalk as cells transition between distinct cellular states. In particular, quiescence in muscle cells is associated with enhanced exosome output and altered exosome cargo; these G₀-derived exosomes exhibit higher capacity for inducing differentiation in MB and greater uptake by MT. In summary, the results suggest that quiescent muscle cells have a greater capacity for signaling cross talk than previously appreciated, by enhancing differentiation of proliferating myoblasts, and/or by reinforcing resilience/differentiation of myotubes (Figure 8).

The following reagents were used; 2, 3-Butanedione monoxime (BDM), (B0753-100G, Sigma), Heparin (Sigma) and Wortmannin (Calbiochem) Dynole (ab120463, Abcam), CHIR99021 (StemRD), Oil Red O (Sigma) 4% Tryphan blue (Invitrogen), Alexa fluor 488 Phalloidin (A12379), BODIPY (D7545, Thermo-Fisher) and DAPI (Invitrogen). Kits: Click-iT EdU Imaging Kit with Alexa Fluor 647 Azide (C10086, Thermo Fisher Scientific), PKH26 Red Fluorescent Cell linker for General Cell Membrane (MIDI26, Sigma-Aldrich), One-Glo Luciferase assay system (Promega). Antibodies for Western blotting were: mouse anti-Alix (ab117600, Abcam), rabbit anti-flotillin 1 (F1180, Sigma), mouse anti-TSG101 (T5701, Sigma), rabbit anti-CD9 (EXOAB-CD9A-1), rabbit anti-Rab27a (17817-1-AP), rabbit anti-Eea1 (2411S), rabbit anti-calnexin (C4731, Sigma), Myogenin (sc12732, Santa Cruz), p130 (SC-317, Santa Cruz), HSP70 (ADI-SPA-820, Enzo), HSP90 (sc-13119, Santa Cruz), Kibra (8,774, CST) and mouse anti-GAPDH (ab8245, Abcam). HRP-conjugated secondary antibodies were from Calbiochem. Details of antibody dilutions for blotting and immunofluorescence are provided in Supplementary Table 2.

C2C12A2, a strictly anchorage-dependent subclone (Sachidanandan et al., 2002) of mouse C2C12 skeletal muscle myoblasts (Yaffe and Saxel, 1977; Blau et al., 1983) were used in all experiments. All media were supplemented with 1% penicillin and streptomycin (Invitrogen). Only cells that had been maintained in strictly subconfluent conditions were used for preparation of frozen stocks; all experiments were performed on cells between passage two to five after revival from a frozen stock, and monitored for appropriate expression of proliferation and differentiation markers. Adherent proliferating myoblast cultures (MB) were maintained in high serum growth medium [GM; high-glucose Dulbecco’s modified Eagle’s medium (DMEM) with 20% fetal bovine serum; Invitrogen]. For induction of quiescence (G₀), proliferating MB were treated with 2,3-Butanedione monoxime (BDM 30 mM) in GM for 24 h as described (Dhawan and Helfman, 2004; Venugopal et al., 2020). For reactivation (R) of G₀ cultures, 24 h after BDM treatment, G₀ cells were induced to re-enter the cell cycle by removing BDM-media and replacing with fresh GM for different times (6, 12 or 24 h). For induction of differentiation (MT), cultures at 80% confluence were incubated in low serum differentiation medium (DM: DMEM with 2% horse serum), replaced every 24 h for 3 days.

Preparation of Exosome Production Medium (EPM): Exosomes were depleted from 20% FBS or 2% Horse serum-containing medium by centrifugation overnight at 110,000 × g, 4°C as described (Thery et al., 2006) and 0.22-µm filter sterilized after adding appropriate supplements for each culture condition. EPM was tested for efficiency of proliferation and differentiation (Supplementary Figure S1F).

Exosomes were isolated as described previously (Thery et al., 2006). In brief, C2C12 were cultured in different states as described above (MB, G₀, R6, 12 or 24 h, or MT day 3) and switched to fresh EPM only for collection of exosomes for 6 h (conditioned medium, CM), following which exosomes were isolated by differential centrifugation as follows: 300 × g for 10 min to remove dead cells, 2000×g for 10 min to remove residual debris and apoptotic bodies, 10,000×g for 40 min to remove microvesicles, 0.22 µm filtration to remove vesicles larger than 200 nm and finally ultracentrifugation for 110,000×g for 1.5 h to collect sEV/exosomes. The resulting pellets were resuspended, washed once in double-filtered PBS (DPBS) and re-pelleted at 110,000×g for 1.5 h in a Beckman Coulter Ultracentrifuge (Optima XPN), with Type 45 Ti fixed angle, and 94 mL tubes (Beckman Coulter). The final “exosome” pellet was resuspended in DPBS, and protein concentration measured by a BCA Protein assay kit (Thermo-Fisher). Exosome pellets were resuspended in DPBS and stored in aliquots at −80°C.

Linear sucrose gradients were prepared using two sucrose solutions corresponding to 60% (w/w) and 5% (w/w) sucrose in 1 × TNE buffer (25 mM Tris–HCl pH 7.5; 0.15 M NaCl; 1 mM EDTA). 6 mL of 60% sucrose solution was placed in the bottom of an ultra-clear ultracentrifuge tube, overlaid with 6 mL of 5% sucrose solution, sealed and allowed to equilibrate overnight at 4°C. 1.5 mL of sucrose solution was carefully removed and 0.5 mL of exosome suspension prepared as above was gently overlaid on the preformed sucrose gradient, and centrifuged to equilibrium (100,000×g for 16–18 h at 4°C using swinging bucket rotor, SW41Ti rotor, Beckman Coulter). 15 fractions of 750 μL each were carefully collected from the top of the tube. 10 μL of each fraction was used to measure the refractive index (Brixxus-CRI375P, Sartorius) and density was determined using the rotor specification for Beckman Coulter. The fractions were individually washed by centrifugation in double-filtered chilled PBS buffer and the final pellet solubilized with 2x Laemmli sample buffer. Individual fractions were displayed using SDS-PAGE, transferred to PVDF and the blots probed as below.

Exosome size distribution and particle number was determined by nanoparticle tracking analysis (NTA) using a NanoSight system NS300 instrument equipped with a 488 nm laser module (Malvern, United Kingdom) and NTA 3.2 analytical software. The NS300 uses Brownian motion of nanoparticles dispersed within a liquid as its principle of operation. With the aid of a magnification microscope, the laser beam traverses the sample chamber identifying the nanoparticles. Each experiment was carried out in triplicate. Exosome samples obtained from equal number of secreting cells over a 6-h collection period were diluted in DPBS (100-fold) appropriately and injected into the sample chamber. The samples were mixed initially to prevent clumping of nanoparticles which is known to affect the scattering patterns. The experimental setup for NTA was performed on the settings of camera level: 12-13, capture duration: 30 s, no. Of scans: 3 with detection threshold of: 3-4. An average of three scans were taken to represent the particle size and number. The videos were processed by the NTA software, version (NanoSight NTA 3.2) and processed for analysis. Each video yielded the mean, mode, and concentration of particles of the diluted sample (1 mL). The generated data were used for statistical analysis.

Exosomes were evaluated morphologically through negative staining (Thery et al., 2006). The exosome pellet derived from equal number of secreting cells were fixed by resuspension in 100 µL of 2% PFA, 5 µL was deposited on Formvar-carbon coated EM grids (Electron Microscopy Sciences) and allowed to adsorb for 20 min in a dry environment. Grids were washed in 100 µL PBS, fixed in 50 µL of 1% glutaraldehyde for 5 min, followed by seven washes in 100 µL changes of distilled water. The grids were placed in a 50 µL drop of 2% uranyl-oxalate solution pH seven for 5 min followed by contrasting and embedding in a mixture of 4% uranyl acetate and 2% methyl cellulose (1:10) for 10 min on ice in dark. Excess fluid was blotted to leave a thin film over the exosome side of the grid, air-dried and observed under the electron microscope (JOEL JEM 2100) at 120 kV and images captured using Gatan Ultrascan CCD camera. Samples were imaged at ×6700 magnification (scale bar 500 nm).

Exosome pellets isolated by ultracentrifugation as above and lysed in SDS Lysis buffer (0.2 M Tris–HCl pH 6.8, 8% SDS, 0.05 M EDTA, 4% 2-mercaptoethanol, 40% glycerol, 0.8% bromophenol blue). 50 μg of each exosome lysate were separated on a 15% SDS-PAGE gel, proteins visualized by staining with Coomassie brilliant blue R250. Each lane was individually sliced into 10 pieces, and each piece individually cut into smaller pieces (1–2 mm) for processing. Proteins were subjected to reduction, alkylation, and in-gel digestion using Trypsin (Shevchenko et al., 1996). Digested peptides were desalted and enriched using Pierce C18 Tips. Eluted peptides were resuspended in 5% (v/v) formic acid and sonicated for 5 min (Sennels et al., 2009). Samples were analyzed on Orbitrap Exploris™ 240 mass spectrometer (Thermo Scientific) coupled to a nanoflow LC system (Easy nLC II, Thermo Scientific). Peptide fractions were loaded onto a PepMap™ RSLC C18 nanocapillary reverse phase HPLC column (75 μm × 25 cm) and separated using a 60 min linear gradient of the organic mobile phase [5% Acetonitrile (ACN) containing 0.1% formic acid and 95% ACN containing 0.1% formic acid]. For identification of peptides the raw data was analyzed on MaxQuant proteomics computational platform (Ver. 1.6.8) (Cox and Mann, 2008) and searched against UniProt amino acid sequence database of Mus musculus (release 2022.04 with 17,131 entries) and a database of known contaminants. MaxQuant LFQ (Label Free Quantification) feature was used to quantify the differences in abundance between the different exosome states. Proteins with peptide count two or higher were selected for further analysis.

Immunoprecipitation of Kibra-gfp was performed according to ChromoTek GFP Trap agarose (gta-20) manufacturer’s protocol. Transfected cells were washed twice with ice-cold PBS and lysed with lysis buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5% NP40, and 0.5 mM EDTA) containing 1X protease and phosphatase inhibitors. 200 μL of lysis buffer was added to 60 mm culture plate and cells were allowed to stand in the lysis buffer for 2 min on ice, after which, cells were scraped and collected in a 1.5 mL microcentrifuge tube. Lysate was sheared through a 2 mL syringe with 18-gauge needle and incubated at 4°C for 30 min. Cell debris was removed by centrifugation at 17,000xg for 10 min at 4°C and 300 µL of dilution buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, and 0.5 mM EDTA) supplemented with 1X protease and phosphatase inhibitors was added to the supernatant. 50 μL lysate was saved as input control. Lysates containing equal amounts of protein were incubated with 25 µL of GFP Trap beads slurry for 2 h at 4°C with gentle rotation. Post incubation, GFP Trap beads were sedimented/collected by centrifugation at 2500 g for 5 min, 4°C. Beads were washed thrice with 500 µL of wash buffer (10 mM Tris/Cl pH 7.5, 150 mM NaCl, 0.5% NP40, and 0.5 mM EDTA) and finally protein was eluted by boiling beads in 2X Laemmli buffer. Immunoprecipitated samples were analysed by Western blotting.

Neutral lipids and lipid droplet (LD) morphology was studied by oil red O (ORO) staining (Mehlem et al., 2013). Cells grown on coverslips were treated with exosomes and fixed in 4% PFA for 15 min. Filtered working ORO solution (0.5% in isopropanol) was added for 15 min, nuclei counterstained with Hoechst 33,342 (Thermo-Fischer), and examined in brightfield at 100X as well as in fluorescence using a Texas red excitation filter (540–580 nm) and UV excitation filter (340–380 nm) on a confocal microscope (Olympus FV3000). Images were processed and analysed using ImageJ. Membranes enriched in ceramides and sphingolipids were visualized by incubating live cells for 30 min at 37°C with red fluorescent BODIPY-ceramide dye (10 µM, final concentration). Cells were fixed with 4% PFA for 20 min, washed with PBS, permeabilized with 0.2% Triton-X 100 for 10 min, washed again with PBS, nuclei counterstained with Hoechst 33,342, and visualised under a Zeiss LSM880 with Airyscan in Super Resolution Mode (zoom: 3.8 of 100X) using excitation/emission wavelength of 589/616 nm (Software: ZEN 2.3 SP1).

Cells were grown on glass coverslips, fixed with 2% PFA for 20 min at RT, permeabilized for 15 min with 0.5% Triton X-100 in PBS, blocked with 2% horse serum in 0.5% Triton X-100–PBS for 1 h, and incubated with primary antibody diluted in blocking buffer at RT for 1 h or overnight at 4°C. Antibody dilutions are given in Supplementary Table 2. The cells were then incubated in the appropriate secondary antibody for 45 min at RT. Nuclei were counterstained with DAPI (1 μg/mL) in PBS for 15 min. Samples were imaged using a confocal microscope (Olympus FV3000). Image intensity was calculated using Fiji (ImageJ) software, and corrected mean intensity (CMI = total intensity of signal—area of signal × mean background signal) was determined for more than 100 nuclei per sample.

To prepare mCherry- and Flag-tagged Rab27a and GFP-tagged Kibra, gene-specific primers (with indicated restriction enzyme site adaptors) were used to amplify full length murine Kibra (3,312 bp) and Rab27a (665 bp) using cDNA from C2C12 and cloned into eGFP-N3 (XhoI, KpnI sites) and pmCherry-C1 (EcoRI, BamHI sites) respectively. CD63-GFP (CYTO120-PA-1) and CD9-RFP (CYTO123-PA-1) plasmids were purchased from System Biosciences (SBI); YFP-Endo plasmid encoding the endosome marker RhoB (Ghoshal et al., 2021) was a kind gift from Suvendra Nath Bhattacharyya. Promoter-luciferase constructs for Myogenin (1621bp) and MCK (3,357 bp) were prepared by amplification from the genomic DNA of C2C12 and cloned into pGL3-basic vector (E1751, Promega) (Primer sequences are given in Supplementary Table 1, Supplementary Material). C2C12 myoblasts plated on cover slips (for imaging) or 12-well plates (for luciferase assays) or 60 mm dishes (for immunoblotting) were transfected as previously described using Lipofectamine LTX reagent (15338100, Invitrogen) (Sebastian et al., 2009). For signaling assays, cells were treated with exosomes (25 μg/mL) for 24 h by addition to the growth medium.

Stable cell lines derived from single clones of TOPflash or FOPflash transfected in C2C12 cells (Subramaniam et al., 2014) were used for testing Wnt-mediated transcriptional activity of exosomes derived from different cellular states. TOP-Flash construct (Tcf/Lef reporter Super 8x), (Veeman et al., 2003), contains multimerized Tcf/Lef-binding sites; specificity is provided by the control FOP-Flash which contains mutated binding sites. For reporter analysis, target cells in 24-well plates at 70% confluency (both TOP and FOP stable cells) were treated with different donor exosomes (25 μg/mL) or Wnt activator, CHIR99021 (3 and 5 μM) for 24 h. Dual luciferase assay was performed as per manufacturer’s protocols (Roche). BCA protein quantification was performed and relative light units (RLU) measured in a luminometer (Enspire, Perkin Elmer) was normalized to total protein in the lysate (RLU per μg protein). PBS without exosomes was used as a negative control, while FOP-flash cells served as control for specificity of Wnt-activated luciferase.

C2C12 cells plated on cover slips were cultured under different conditions [MB or G₀] were pulsed with 10 µm EdU (5-ethynyl-2′-deoxyuridine) for 30 min prior to fixation with 2% PFA for 20 min, permeabilised and blocked (0.5% Triton X-100 + 10% FBS in PBS). Click-iT EdU reaction cocktail was used as per the manufacturer’s instructions to detect EdU-positive S-phase nuclei and counterstained with DAPI to visualise all nuclei using a Zeiss Axio Imager two epi-fluorescence microscope. No EdU control was negative and no cross reactivity of secondary reagents was detected.

MB cells treated with exosomes derived from different donor cells MB, G₀, R24 and MT (25 µg) for 24 h, were plated at clonal density (500 cells/150 mm dish) and cultured for 7 days. Colonies were stained with methylene blue for counting.

For all experiments unless otherwise mentioned, three biological replicates were used. Statistical analysis values are presented as mean ± SEM. Statistical differences between means were determined using the unpaired/paired Student’s t-test or one-way ANOVA as indicated. All analyses were computed using GraphPad Prism six software. Differences were considered as statistically significant for p < 0.05. Digital images were captured and processed using ImageJ (Fuji); composites and overlays were prepared using ImageJ; minimal adjustments to brightness or contrast were uniformly applied to entire image. The spatial colocalization between different proteins of interest each labelled with a different fluorophore was measured by calculating Pearson’s correlation coefficient (r) using the JACoP plug-in of Fiji (ImageJ) software.